DC coupled Sand 'n Glass with Current Feedback

[ChocoHolic]

Hi Tube Guys !
... obviously I am not much into tubes, but I always wanted to start playing with glass.
By chance I would need a gain stage + Buffer with high input impedance and the possibility to have inverting/non inverting inputs.
This seems to be a good starting point for me, because tubes offer high input impedance and reasonably nice noise figures at the same time.

Played around with simulations and this circuit seems to be promising...
Feel free to jump in and rip it apart

[jan.didden]

Can't see anything obviously wrong with it

Anode resistors seem low; what's the OK gain? Don't you want currents sources in the anode leads?

jan didden

[DF96]

Designed for maximum 'tube sound' i.e. lots of low order distortion?

[ChocoHolic]

Yes the anode resistors are low.
I have to run from low rails and still want to allow output levels up to +/-40V.
Furtheron the drive impedance for the BJTs appeared fine to me.
Last but not least, I want to keep it simple.
Of course this is not a natural match to most triodes, agreed.
But the 6DJ8 seemed to be a reasonable fit.
Which gain do you mean by 'OK gain'?
Open loop, Anode sloping divided by input sloping?

Here two pics from the proto.
Unfortunately I do not have ECC88 or 6DJ8 on hand.
For today I decided to play with an old ECC82 that I found in my playground.
From data sheets ECC82 is not reasonable here, but I simply could not wait until the 6DJ8 will arrive...

[SY]

If I were to build such a thing, I'd look into using pentodes instead of triodes.

[ChocoHolic]

Originally Posted by DF96

Designed for maximum 'tube sound' i.e. lots of low order distortion?

In simulation the spectrum appeared to be reasonable.
In reality with the preliminary set up with the ECC82 the distortions are not bad.

Funny thing is that that at +/-55V K2 does almost vanish, while at +/-50V and +/-60V it shows similar or even higher levels as K3.
Here the measurements with an output swing of 5Vp.

[ChocoHolic]

Hm... at +/-55V not just K2 does vanish, but also K3 is pretty low.

In fact with that operating point on the ECC82 I am already in the range of positive Vg.
At Vg around 700mV there seems to be a minimum in distortion.
I think there was an old IEC paper on that..., but also read somewhere here in forum that such operating point would lead to unpleasant higher order distortion.
So far my measurement does not show this effect.

[DF96]

The triodes will create distortion with such low anode loads, but the NFB will reduce it again. Triodes are only linear with a high impedance load.

You might have problems with the output voltage bottoming on either the BJT collector or the CCS. This is because you have, in effect, two CCS both trying to control the same current. The BJT current is set by some resistors and the LTP tail CCS. Might not show up in simulation, where everything is perfect.

[jan.didden]

Originally Posted by ChocoHolic

Which gain do you mean by 'OK gain'?
Open loop, Anode sloping divided by input sloping?

Sorry, typo - I meant OL gain...

[ChocoHolic]

Hi DF!
...., you mean I should spend the efforts to run from higher rails and put current sources to the anodes?

So far the circuit does not show any fundamental issues, the two CC tails do not fight each other. Please note all shown measurements in this thread are from the real proto, including distortion measurements.
Furtheron with the ECC82 there is not much loop gain to reduce distortions by feedback.

Hi SY!
Pentodes. Erhm, ... you know... I did not dare to start my glass excursion with pentodes.


...and here a measurement on the step response, showing a +/-6V rectangle with 20kHz. (Probes: 10X).